Abstract: Disclosed are a net load forecasting method and apparatus for a new energy electric power market. The method includes: obtaining and performing data preprocessing on new energy output data and external environmental data, and extracting strongly correlated features from the new energy output data and the external environmental data after the data preprocessing; performing feature expansion on the strongly correlated features, and inputting the strongly correlated features after the feature expansion into a preconstructed regression forecasting model, to obtain a first forecast value; obtaining and performing data preprocessing on user load data and load influencing factor data, and inputting the user load data and the load influencing factor data after the data preprocessing into a FNN-LSTM hybrid model, to obtain a second forecast value; and calculating a difference between the second forecast value and the first forecast value, to obtain a net load forecasting result.

Abstract: Disclosed are a net load forecasting method and apparatus for a new energy electric power market. The method includes: obtaining and performing data preprocessing on new energy output data and external environmental data, and extracting strongly correlated features from the new energy output data and the external environmental data after the data preprocessing; performing feature expansion on the strongly correlated features, and inputting the strongly correlated features after the feature expansion into a preconstructed regression forecasting model, to obtain a first forecast value; obtaining and performing data preprocessing on user load data and load influencing factor data, and inputting the user load data and the load influencing factor data after the data preprocessing into a FNN-LSTM hybrid model, to obtain a second forecast value; and calculating a difference between the second forecast value and the first forecast value, to obtain a net load forecasting result.